Continuity and Open Threads
This chapter collects every point on which the fourteen volumes, the two companions or the dossier's own chapters disagree, leave something undefined or stop short of an outcome. Each of Chapters 3 to 26 ends its subject with an open-thread note; the extracts of data/extracts/ carry the same findings in section 8 of each file. Here they are merged, de-duplicated and classified, so that the author can work through them in one place. The findings are for the author. The dossier does not silently choose: where a chapter needs one value to stay readable it uses a working value, says so, and the working value is recorded below beside the recommended resolution. The chapter draws on all fourteen volumes and on Landscape and Tools, and it is the record that Chapter 28 refers to when it asks what has to be fixed in the fiction (2, Nordqvist, 2026).
How the register is built
Each entry has an identifier OT-HM-nn, a statement of the thread, the volumes and chapters that hold each side, the working value this dossier uses and a recommended resolution. The entries are grouped by kind and, within a kind, the most consequential come first. Five kinds are used.
| Kind | Meaning |
|---|---|
| Contradiction | Two or more volumes state incompatible values for the same fact: a count, a level, a classification, a rule. |
| Identifier collision | One identifier is used for two things, or one thing carries two identifiers or two spellings. |
| Date clash | Two volumes, or a volume and its own identifier, date the same event differently, or a date falls where the rule forbids it. |
| Missing outcome | A test, review, plan or decision is stated and its result is never recorded, or a thing is referred to and never defined. |
| Scope mismatch | Two statements are both right for their own scope, but the scopes differ (a unit, a market, a clock, a time), so a reader who combines them is misled. |
The weight given to an entry is the dossier's own judgement: an entry is consequential when it changes what a reader would conclude about how far an agent is trusted, who answers for it, or how old the evidence is. Two threads head the register for that reason, the number of Mode 5 agents and the level of the security-operations agent HM-SEC-001, and the incident identifiers of 2027 head the identifier group. The status of all values is the status of their volumes: the organisation's profile to 16 November 2026 is established, and the later years are illustrative fiction written in November 2026 (2, Nordqvist, 2026).

Figure 27.1: The register by kind, and how a finding reaches the author.
Contradictions
| ID | Thread | Held by | Working value | Recommended resolution |
|---|---|---|---|---|
| OT-HM-01 | Number of Mode 5 agents | Two: profile, AI Governance, AI Threat Ch. 1 and 33. Three: AI Red Team; The Transformed Security Architect Ch. 25. Four: that volume's Ch. 8; Tools Ch. 20 (CT, ITO, SOC, EVC); Tools Ch. 18 lists five domains. Six: Landscape (HM-CT, ITO, SOC, CART, EVC, AB) | Six | Define "Mode 5 agent" once (own identity, write access, no human in each action), publish the count with the definition, and update the profile |
| OT-HM-02 | Level of HM-SEC-001 |
Level 2 Supervised: AI Red Team Ch. 4, 5, 9, 17; AI Threat Ch. 18 and 33; Tools Ch. 12. Level 3 for two actions: AIDR-0311 (proposed in Ch. 6), Ch. 18, 21, 27; Landscape. Three actions: AI Blue Ch. 13. Five delegations: AI Blue Ch. 4. Two: AI Blue Ch. 1 |
Level 3 for two reversible actions | Give one dated sequence (start at Level 2, AIDR-0311 approved on a date, third action added on a date) and state the approval of AIDR-0311 |
| OT-HM-03 | Number and mode of SOC and loop agents | AI Red Team: one (HM-SEC-001). AI Blue: five agents, HM-SEC-002 Mode 5 identity at Level 1 (Mode 4 Level 1 Assisted in AI Red Team), HM-SEC-005 Mode 4. AI Purple: profile says four (MND-HM-EMU-01 to -04, "indicative"), Ch. 14 fixes seven mandates and eight functions |
Five SOC agents; seven loop mandates | Replace the profile's "four" with seven; list each agent once with mode and level |
| OT-HM-04 | Stale-terms incident | INC-2026-0457 (37 answers, 72 hours, Sweden): AI Governance Ch. 11. INC-2026-118 (41 hours, 312 drafts): Ch. 16. "212 clauses": Ch. 23 |
Three accounts of possibly one event, not merged | Say whether it is one incident, and give it one identifier |
| OT-HM-05 | Risk class of the triage agent | Limited risk (AIDR-0024 configuration, AI Platform and Security Ch. 11); "not high-risk" (Ch. 23); high-risk for allocation from AIDR-0651 (AI Boss Ch. 6); highRisk: false in an incident record (AI Red Team Ch. 20) |
Limited for triage; high-risk for the allocation function from 2026-12-14 | State that the system has two functions with two classes, and align the incident record |
| OT-HM-06 | Exit plans and drills | AI Governance Ch. 6: EXIT-004 (hyperscaler EU region, open-weight second source at 0.89). Ch. 7: strategy for "Model Provider Alpha", rehearsed 2026-06-12. Drills of 9 and 4 hours against a 48-hour switch target (AI Platform and Security Ch. 25) |
Both kept; no link drawn | Say which plan EXIT-004 implements and how the 48-hour target relates to the 4-hour drill |
| OT-HM-07 | Level of restore-overwrite | Level 1 Assisted: profile, AI Operations Ch. 10 and 30. "Nothing delegated": Ch. 16 and 29 | Level 1 Assisted | Decide whether "Level 1 Assisted" and "not delegated" are the same; use one phrase |
| OT-HM-08 | Level 4 in practice | AI Boss Ch. 8: sales score "Level 4 in all but name" in one unit; Ch. 31: "by accident"; Ch. 30: no action of any system ran at Level 4 over a person | The record: Level 0 for the sales score | State that the practice exceeded the recorded level, or withdraw the phrase |
| OT-HM-09 | Ceiling for decisions about people | DRF-HM-01 (AI Boss Appendix A): "any: L2_Supervised". Ch. 30, 32 and the plan: no system above Level 1 Assisted for M3 Consequence |
Level 1 Assisted | Correct the file or state why it is looser than the rule |
| OT-HM-10 | Ceilings of DLG-HM-EXE-01 |
Profile: strategic, resource, structural, risk, crisis at Level 0, external at Level 1. AI Leader Ch. 15 table: strategic 1, resource 1, structural 0, risk 0 (highest levels per class). Ch. 8: "two routine classes" at Level 3 | Profile | Say that the Ch. 15 table shows highest levels reached, not the ceiling |
| OT-HM-11 | Builder lane ceiling | Vibe to Dev Ch. 21: Level 3 Bounded in the lane. AI Developer: one Level 3 class, patch-level dependency updates (DLG-HM-DEV-01) |
Both | Say whether the lane is a second Level 3 class and add it to the delegation profile |
| OT-HM-12 | Reporting per named person | AI Developer: KBV-IT-2024-DEV-01 bars reports per named developer. Vibe to Dev Ch. 19: lane logs identify named builders (works council informed) |
Both | Add the lane to the agreement or anonymise the logs |
| OT-HM-13 | Applicability of NIS 2 | Displaced for EU financial entities: AI Leader Ch. 24 (Article 4(1), BSIG section 28(6)). Depends on national transposition: AI Operations Ch. 2; AI Blue Ch. 19 | Both, with the Chapter 4 reading | Have the author or counsel settle it once and align three volumes |
| OT-HM-14 | Location of the DORA flag | dora: law in the group charter (AI Governance); in each market overlay (AI Platform and Security Ch. 24) |
Overlay, as the later volume | State that the overlay holds the flag and the charter the commitment |
| OT-HM-15 | Token life | 5 minutes: AI Red Team Ch. 4; AI Threat Ch. 24; CP-HM-ITOPS-01. 15 minutes: AI Red Team Ch. 13, 14, 18. 15 cut to 5 after the first drill: The Transformed Security Architect |
15 before the first drill, 5 after | Give the date of the cut and correct the 5-minute statements that precede it |
| OT-HM-16 | Level 3 rate limits | 5 targets an hour (Ch. 5), 5 workstations (Ch. 6), 10 actions (Ch. 18), 20 in the containment catalogue (Ch. 21), all AI Red Team | Five per hour at the policy point | Say which limit is the policy and which the catalogue ceiling |
| OT-HM-17 | Stop targets and defaults | 120 seconds (group default, HM-SEC-001); 60 seconds (loop runner; AI Purple Ch. 4 and 8 first said 120). Tools Ch. 25: 30 minutes, 5 minutes, 120 seconds, 60 seconds |
120 and 60 as stated | List each default with its object |
| OT-HM-18 | Holders of the stop record | Landscape Ch. 13: SOC duty analyst on call and system owner. Ch. 17: agent operator on call and head of group AI platform. Budget of five minutes in both | Both recorded | Pick one pair of holders for the 2027 drill |
| OT-HM-19 | The board's act of 31 March 2027 | Profile and The Transformed Security Architect: asks for time to stop. Landscape Ch. 13 and 18: takes the decision right, tier 2 to tier 4 for HM-ITO only | Both, in that order | Say whether HM-CT and HM-SOC are also at tier 4 |
| OT-HM-20 | Owner of the IT-operations agent | Site-reliability team, Head of IT Operations, head of group AI platform, executive board at tier 4 (several volumes) | Four kinds of ownership: manifest, residual risk, seat, decision right | State the four in one table |
| OT-HM-21 | Claims assistant | Limited risk but excludes life and health; languages German, English, Swedish, Danish (AI Platform and Security Ch. 10; no Danish market), five (AI Governance Ch. 4), six markets; evaluation slices de and sv; pilot about 320 handlers against about 180 (UC-2026-052) |
Five languages; 320 handlers | Remove Danish or add a market; say that 180 belongs to another system |
| OT-HM-22 | Group IT headcount | AI Operations: about 1,300 (parts sum to about 965). AI Developer: Application Development about 480, same CIO | Not reconciled | State whether the 480 are inside the 1,300 |
| OT-HM-23 | The COO seat | AI Governance board has a COO; AI Leader executive board lists none and titles the CIO chief information and operations officer | Same seat assumed | Say so |
| OT-HM-24 | UK and Swiss entities and the AI Act | No duty for the UK and Swiss entities, yet Hansa is provider of the underwriting-support model (AI Red Team Ch. 21); the Act can reach output used in the Union (AI Blue) | Group applies the baseline there as policy | Add the output-used-in-the-Union qualification to Ch. 21 |
| OT-HM-25 | Retention of employment records | Six months after rejection (AI Boss Ch. 7, 25); 24 months for MDR-HM-0001 under AIDR-0679 |
Both | Say which applies to which record |
| OT-HM-26 | Polish Labour Code | Articles 22 squared and 22 cubed (AI Boss Ch. 6); Article 22 cubed (profile) | Both articles | Verify against the text |
| OT-HM-27 | Evidence rung of THR-HM-03 |
Laboratory demonstration with label observable (AI Threat Ch. 22, 32); incident in the wild with observable (Ch. 5, CAP-02) | Register | Explain the divergence as Ch. 32 does for THR-HM-12 and -13 |
| OT-HM-28 | Illustrative decisions used as real | AIDR-0960 and AIDR-0970 (restart): illustrative in the ledger and AI Operations Ch. 17, used as actual in Ch. 29 to 31 and RBK-HM-0007 |
Treated as illustrative | Remove the label or the use |
| OT-HM-29 | Operations agents and restart | Profile lists HM-OPS-001 to -004; AI Operations Ch. 16 adds -005, -006. Level 3 for restart "already exists" (Ch. 4) but holds only on condition of OBS-HM-0001 and used 73 per cent of the budget (Ch. 29) |
Six agents; conditional Level 3 | Update the profile; reword Ch. 4 |
| OT-HM-30 | Screening and TalentLens |
Profile: people Level 1 for screening, existing. AI Leader Ch. 4: screening "sits outside" the board's profile; no chapter of its own | Level 1 | Give TalentLens a place in the delegation profile |
| OT-HM-31 | Supplier identity | SXR-HM-2026-03 "Provider A"; "Model provider A" under the triage agent (AI Threat Ch. 29); a US provider for classification (Ch. 15); "Model Provider Alpha" (AI Governance Ch. 7) |
Not assumed equal | Say whether A and Alpha are one supplier |
| OT-HM-32 | Hosting of HM-CT | Landscape: own premises. Profile: works through the Polish centre so that calls bind to a market | Hosting versus operation | Add both |
| OT-HM-76 | Measurement base of September 2032 | Chapters 2, 11 and 22 of AI Environment: platform racks 62 MWh (85.6 MWh at facility, 30.0 t); Chapters 8 and 18 and FPR-HM-0002: GPU racks 49,800 kWh; about 3.4 million requests "across all its uses" against about 1.5 million in the breakdown. The claims assistant's 0.996 Wh (node), 3.91 and 2.42 Wh (rack IT) and 0.30 Wh (facility estimate) are at different boundaries and are never added |
Both, labelled by boundary | Explain the 62,000 minus 49,800 kWh and the request count |
| OT-HM-78 | German PUE limits | Chapter 18 of AI Environment: 1.5 from 2027-07-01 and 1.3 from 2030-07-01 in the statute, 1.4 in the bill; Chapter 21: bill 1.6 from 2027 and 1.4 from 2030 (a margin of 0.02 against 1.38); the bill is speculation | Text in force as given by Book 14 (statute not read); bill not used | Align the two chapters; the lawyers read the text in force on the date of decision |
| OT-HM-79 | Routing and energy policy RP-HM-CLAIMS-2026-02 |
AI Threat: monthly energy budget 40 MWh, share trigger below 0.7, at most 500 output tokens, cache life 24 hours; AI Environment: revision 3 of 2032-11-02 as a cascade (22 per cent escalation, alarm 0.35, freeze 0.50, break-even 0.67), budget EBG-HM-0002 of 1,260 kWh a month before facility overhead |
Each as its volume states; revision 3 is later | Record the lineage of the revisions |
| OT-HM-84 | Owner of FPR-HM-0002 |
AI Environment tables: Katrin Ohlsen; its Appendix A record: anja.weiss; the profile says business owner Ohlsen, producer Weiß, accountable Dr. Verena Lindner |
Three roles, as the profile | Name one owner in the record |
| OT-HM-86 | Wording of BetrVG section 87(1) no. 6 | AI Boss and the earlier text of Landscape: technical equipment suited to monitor behaviour or performance; Landscape now: technical devices designed to monitor, objective suitability enough whatever the employer's intention, citing the Bundesarbeitsgericht (2024); the statute says "bestimmt sind" (designed); AI Environment names no works agreement and relies on section 87(1) no. 6 and Thomas Brandt | The statute's word, "designed", with Landscape's gloss | Reword AI Boss against the statute and the judgment |
| OT-HM-88 | Works-council rows by country | AI Boss: Netherlands WOR art. 27(1)(k) and (l); Sweden unions under MBL section 11; Landscape now: Netherlands art. 27(1)(l), Sweden has no works council of this kind (marked for review in its fact-check) | Both; Sweden as unions under the Co-Determination Act | Check the Dutch points and the Swedish wording |
(2, Nordqvist, 2026) (3, Nordqvist, 2026) (4, Nordqvist, 2026) (5, Nordqvist, 2026) (6, Nordqvist, 2026) (7, Nordqvist, 2026) (8, Nordqvist, 2026) (9, Nordqvist, 2026) (10, Nordqvist, 2026) (11, Nordqvist, 2026) (12, Nordqvist, 2026) (13, Nordqvist, 2026) (15, Nordqvist, 2026) (Tools, .) (16, Nordqvist, 2026) (14, Nordqvist, 2026) (15, Nordqvist, 2026)
Identifier collisions
| ID | Thread | Held by | Working value | Recommended resolution |
|---|---|---|---|---|
| OT-HM-33 | Incident identifiers of 2027-03-09 | HM-INC-2027-0142 (AI Red Team Ch. 20, 23), INC-HM-2027-0412 (Ch. 21), INC-HM-2027-0533 (Ch. 22); the first two coincide at 02:41+01:00 and 02:40Z. AI Blue separates INC-HM-2027-0412 from INC-HM-2027-0455 after an editorial change |
Three distinct events, and 0455 a fourth |
Adopt the INC-HM-yyyy-nnnn format and say whether the first two are one event |
| OT-HM-34 | RST-HM-2031-0001 |
Revert rehearsal of 2031-01-08, derived (AI Developer Ch. 31); restore test of 2031-11-14 (Vibe to Dev Ch. 14); exit test is RST-HM-2031-0002 |
Collision, unresolved | Renumber one |
| OT-HM-35 | HM-SEC-006 |
Continuous security chaos engineering (AI Red Team); alert-triage agent (AI Blue) | Both | Renumber one |
| OT-HM-36 | Findings and regression cases | AI Purple: FND-HM-2028-003 used twice (now -001 for the SOC agent), FND-HM-2028-031 became -012, REG-HM-2028-009 became REG-HM-0032; the decisions of 2028-04-14 (AIDR-0888 to -0890) cite AP-HM-LOOP-03 and -04 in Ch. 27 |
The ledger | Align the chapters with the ledger |
| OT-HM-37 | AIDR numbering |
AIDR-0685 was AIDR-0670 and AIDR-0653 to -0657 were -0670 and -0671 (AI Boss). UC-2026-031 is AIDR-0147 in AI Governance Ch. 16, none in the profile; AIDR-0063 and -0034 appear only in the profile; AIDR-0027 is the IT-operations agent in the profile and a recommend-only pilot in Ch. 23 |
The profile | Publish one table of use cases and their records |
| OT-HM-38 | Spelling variants | DLG-HM-OPS-001 and -02; DLG-HM-EMU-01 and DLG-HM-LOOP-01; REV-HM-0014 and REV-HM-0001; OQM-HM-CLM-01 in two forms; CNT-HM-0001 with two holders; DEC-2026-031 and DEC-HM-004 (health underwriting triage assistant against underwriting-support model); DET-HM-0455 with differing dates; MND-HM-01 written version 3 and 4 |
Both spellings kept | Choose one form per family |
| OT-HM-39 | The works agreement | KBV-IT-2024-DE (Hamburg) (Vibe to Dev); KBV-IT-2024 with annexes and draft KBV-IT-2024-DEV-01 (AI Developer) |
One agreement named three ways | Name it once |
| OT-HM-40 | Names of an action and of an owner | disable-session-token (AI Red Team Ch. 5) against revoke-user-sessions and idp.revoke-sessions; owner group-soc-lead (Ch. 9), head-of-group-soc (Ch. 6) |
The manifest names | Use the manifest names throughout |
| OT-HM-83 | Renumbering inside AI Environment | EBG-HM-0002 (Chapter 13 became EBG-HM-0003, Chapter 17 EBG-HM-0004, Chapter 18 EBG-HM-0005, Chapter 19 EBG-HM-0006); SDR-HM-0002 renumbered in Chapter 23; AIDR-1351 became AIDR-1366 and AIDR-1347 and AIDR-1397 became AIDR-1329 and AIDR-1379; the blueprint of Book 14 is generated from outline data and still carries older references |
The numbers of the register and Appendix A of Book 14 | Regenerate the blueprint |
(4, Nordqvist, 2026) (7, Nordqvist, 2026) (8, Nordqvist, 2026) (9, Nordqvist, 2026) (12, Nordqvist, 2026) (13, Nordqvist, 2026) (Tools, .) (16, Nordqvist, 2026) (14, Nordqvist, 2026)
Date clashes
| ID | Thread | Held by | Working value | Recommended resolution |
|---|---|---|---|---|
| OT-HM-41 | First stop drill | SPT-HM-ITOPS in Q4 2026, failed, no retest (AI Threat Ch. 24); STOP-HM-ITOPS-2027-Q1-01, 17 minutes against five, placed in "quarter two" and already failed in the report of 2027-03-31 (The Transformed Security Architect Ch. 25) |
Q1 2027, reported 2027-03-31 | Say whether the two drills are one |
| OT-HM-42 | Triage go-live | Conditional go-live of UC-2026-014 on 2026-10-21, start 2026-11-04, as an assistant (AI Platform and Security); spend incident SPD-2026-007 on 2026-09-22 and a 2027-Q1 forecast for the agent (AI Governance Ch. 16A) |
A pilot scope of a system already running as an agent | State the sequence of assistant and agent |
| OT-HM-43 | Allocation reclassification | AIDR-0651, 2026-12-14, release 7.0; 2027-04-30 in Landscape, where it is the first report under AIDR-0650 |
Decision 12-14, report and seat change 04-30 | Align Landscape |
| OT-HM-44 | Start of HM-AB | 2027-01-01 (model); "from January 2027" and "second half of January" (profile) | Set up 01-01, agents late January | State both events |
| OT-HM-45 | Waiver length | WAIVER-2026-007 runs 92 days (2026-08-20 to 2026-11-20); the limit in AI Governance Ch. 13 is 90 |
92 recorded | Shorten or add the exception |
| OT-HM-46 | Notice under Article 26(7) | TalentLens notice issued 2026-06-01; the article applies from 2027-12-02 after Regulation (EU) 2026/1744 (AI Boss Ch. 28 explains) |
Early notice | State the date of application beside the notice |
| OT-HM-47 | Consultation status | AI Governance Ch. 12: Germany, Sweden, UK done, Netherlands in progress, Norway and Switzerland not started; plan in Ch. 23 completes Netherlands, Norway, Switzerland | Both | Date the status |
| OT-HM-48 | Hunt timing | Two cycles 2027-02-01 to 2027-04-30 with a test passed on 2027-04-02 (profile); hunt starts 2027-02-15, repeated by 2027-08-31 (AI Blue Ch. 26) | Profile | Align |
| OT-HM-49 | Register review date | Review of 2026-10-30 precedes AIDR-0470 (2026-12-14) and the snapshot of 2026-11-16 (AI Threat Ch. 32) |
Register | Re-date the review |
| OT-HM-50 | Builder dates | Tier decision 2031-09-18, screening gate 2031-10-20, "starts at Tier C" 2031-09-15; R2 due 2032-01-13 against a plan starting 2032-02-02 (Vibe to Dev Ch. 2, 14, 16, 24) |
The records | Align text with records; say why R2 is open |
| OT-HM-51 | Dates of leadership and developer records | BRD-HM-001 approved 2029-11-12, noted 2029-11-26; AIDR-1051 dated March 2029, filed after AIDR-1050; AIDR-1053 2029-10-15 against a proposal on 10-08 and resolution on 10-09; ROLE-HM-DEV-01 (AIDR-1164) undated; AIDR-1152 dated only 2030; review of Sunday 2031-03-30 held on 2031-03-28 |
The ledger | Date each record |
| OT-HM-52 | Clock of RST-HM-2027-007 |
Expired 2028-04-02 (AI Operations), yet the cooling failure of 2027-11-16 (INC-HM-2027-0961, 1 hour 35 minutes) is treated as evidence from operation |
Expired | Decide, with the head of operational resilience, under DORA Article 11(6)(a) |
| OT-HM-82 | Revisions of the claim CLM-HM-0001 |
Revision 2 is "of 2032-12-06" and approved 2032-12-13 (AIDR-1372); a table of Chapter 27 of AI Environment says that revision 3 needs a production reading, and no revision 3 is written; revision 2 lapsed 2033-01-26 and AIDR-1377 (2033-05-10) did not renew it |
Revision 2 only | Say whether revision 3 exists |
(2, Nordqvist, 2026) (3, Nordqvist, 2026) (5, Nordqvist, 2026) (6, Nordqvist, 2026) (7, Nordqvist, 2026) (8, Nordqvist, 2026) (10, Nordqvist, 2026) (11, Nordqvist, 2026) (12, Nordqvist, 2026) (13, Nordqvist, 2026) (Landscape, .) (15, Nordqvist, 2026) (14, Nordqvist, 2026)
Missing outcomes
| ID | Thread | Held by | Working value | Recommended resolution |
|---|---|---|---|---|
| OT-HM-53 | Stop-path tests | Called quarterly; one record (Q4 2026, fail), no retest (AI Threat Ch. 24); RBK-HM-0007 points to STOP-HM-ITOPS without a later drill (Tools Ch. 25) |
No retest assumed | Record a retest and its result |
| OT-HM-54 | Exit evidence | RST-HM-2031-0002 passed 2031-01-19, evidence ended 2031-07-18 |
Lapsed | Record a retest |
| OT-HM-55 | Open at the report date, AI Purple | IND-HM-2028-Q1 (expires 2028-05-22), COV-HM-0112 (2028-05-09), FND-HM-2028-003 and -004, Swiss audit-log decision (accepted to 2028-06-14), T1213, T1550.001, T1486 and T1490 never run, no S4 scenario against staff, next TLPT, AIDR-0885 conditions; held-out share 13 of 87 (14.9 per cent, under the 15 per cent rule), extension to 20 undecided |
Open | Record the outcomes |
| OT-HM-56 | Open targets, AI Blue | Reproduction 0.80 missed; unassisted 0.10 against 0.20; H1 gate open; mandate review of HM-SEC-001 (expires 2028-02-09) has no AIDR |
Open | Number the review |
| OT-HM-57 | Open gaps, AI Threat | Second source not equivalent for complex claims routes (THR-HM-06 to -08); accumulation model rests on small loss history; Level 3 for HM-SEC-001; Norwegian EEA timing |
Open | Record the decisions |
| OT-HM-58 | Unrecorded results, AI Leader | Rerun of ACC-HM-2029-01, summaries retest, resampling, seeded-error exercise; ADP-HM-2029-01 expired 2029-11-30 and was shown as expired; SPN-HM-2029-01 "will be written"; tripwire, pause exercise and dashboards are "in this illustration" |
Illustrative | Record results or label the illustration |
| OT-HM-59 | Unrecorded outcomes, AI Developer | Review of MND-HM-DEV-02 (2031-01-07); stop-path exercise (2031-01-12); error-budget pilot (AIDR-1166, review 2031-04-20); HM-DEV-003; class renewal to 2031-09-30 with no count of merged or reverted updates; one week at Level 2, restored 2031-01-08 |
Open | Record outcomes |
| OT-HM-60 | Board figure and register | The 42 per cent not confirmed corrected in board minutes; DORA class of the nearshore partner; Alster's assistance a blind spot; register reread of 2031-07-28 without outcome; works agreement KBV-LDR-2030-01 and six other arrangements (check by Zimmermann, 2031-01-31); supervisory board composition and legal basis flagged for verification |
Open | Confirm |
| OT-HM-61 | Register of information | Own entry for the builder lane's assistant (Vibe to Dev Ch. 19) and for the nearshore partner; missing notice clause for Mode 0 at the service desk (AI Platform and Security Ch. 24) | Open | Decide |
| OT-HM-62 | Hosted prototype | About 40 rows of real data pasted into a hosted builder with an open link; data protection officer informed; no incident record, breach assessment or access log; tool without contract (Vibe to Dev Ch. 2, 20) | No record | Open an incident record |
| OT-HM-63 | Second market and works councils | Vibe to Dev Ch. 24 asks for written notice before a second market; Polish and Dutch representatives have already asked for the same terms and an annex is unsigned (AI Developer) | Both | Sign or schedule |
| OT-HM-64 | Unresolved by AI Boss | Whether the AI boss leaves Germany (an EWC question); one system with seven condition sets or seven; who becomes team leader when allocation is learned through 10 per cent of work; oversight reported only at department level, Jana Petersen the only team leader | Open | Decide |
| OT-HM-65 | Plans and baselines | Plan of The Transformed Security Architect Ch. 26 (second reviewer pool, Poland in months 12 to 18, four AI-03 architects, YAML with M1 to M3 only); UK logging exception outcome; September baseline of 6.8 days did not exist when the intent was written; only comparison is before and after (Vibe to Dev) | Open | Complete |
| OT-HM-66 | Referenced, not defined | BTP-HM-2027, format of SDF-HM-2028-002, HM-COMMS-RC-*, purple-align; author's notes missing in Landscape |
Not used | Define or remove |
| OT-HM-85 | Open at the end of the year of AI Environment | Cloud A burst unmetered (rung 3 for energy until 2034); water per use in Frankfurt; the district-heating study (AIDR-1368); the metered reading of AIDR-1379; a production-boundary reading for FPR-HM-0001 from tagged rack meters; Cloud B's basin (so PLC-HM-0001 blocks a move there); no benefit record for the engineers' tools; the UK and Swiss regimes not examined; which figures the sustainability statement carries; the Taxonomy indicators for insurers not read; FPR-HM-0004 and -0005 not written |
Open | Record the outcomes in the next volume |
(5, Nordqvist, 2026) (6, Nordqvist, 2026) (9, Nordqvist, 2026) (11, Nordqvist, 2026) (12, Nordqvist, 2026) (13, Nordqvist, 2026) (15, Nordqvist, 2026) (Tools, .) (16, Nordqvist, 2026) (14, Nordqvist, 2026)
Scope mismatches
| ID | Thread | Held by | Working value | Recommended resolution |
|---|---|---|---|---|
| OT-HM-67 | Entities the model does not draw | UK, Swiss and Polish entities are suppliers in Landscape, but no unit refers to them, so V3 draws no container; the profile has UK, Swiss and Norwegian entities apply AI Act controls as policy; levels per market for the operations agent not specified (AI Operations) | Profile | Add per-market marks |
| OT-HM-68 | Units against systems | HM-TL (M3, Level 1) and the allocation function of HM-CT are separate units; HM-CA reach L against about 320 handlers; HM-CT XL without a count; the "claims triage assistant" in Tools Ch. 3 is the predecessor UC-2026-014 (AIDR-0063), not HM-CT |
Landscape | State the mapping |
| OT-HM-69 | Evidence clocks | 90, 180 and 365 days; valid_days 365 against 90 for the takeover test; CMP-HM-TL-014 90 days against 12 to 24 months; coverage vocabularies of three and six statuses (Tools Ch. 25) |
Each volume's own | Publish a table of clocks |
| OT-HM-70 | Counts that are series | Verified scenarios 25, 23, 27 are a time series, with 25 as baseline in Ch. 17 and 25; not-verified counts 9, 11 and 7 at three dates (AI Purple) | Series | Label each with a date |
| OT-HM-71 | Median approval time | 28 seconds (The Transformed Security Architect Ch. 24); under 30 seconds (Ch. 14); "seconds" (Ch. 25); 2026 Q3; approvals under four seconds flagged (AI Governance Ch. 25) | 28 seconds | Say what each measures |
| OT-HM-72 | Roles named late | CIO and CRO unnamed in AI Operations, named in AI Leader (Rainer Lehmann, Dr. Annette Krämer); the head of operational resilience is named in neither; head of claims operations Joanna Kaczmarek (2031) against chief claims officer Katrin Ohlsen (2029); Katrin Wendt "the only" committee chair | AI Leader | Confirm each holder |
| OT-HM-73 | Shared figures with Nordvik Region | EUR 18,000 a month for the route budget and about 40,000 claims a month equal Nordvik Region's figures; EUR 5,600 for 14 seats and the builder spend alert at 80 per cent are illustrative in their volumes | Coincidence recorded | Change one set |
| OT-HM-74 | Plan wording | AI Boss Ch. 32 describes the plan in prose, not all of H1 to H6; Appendix A and the profile list H1 to H6 | H1 to H6 | Align |
| OT-HM-75 | Footprint of the claims-triage agent | FPR-HM-0002: 2.3 Wh bottom-up against about 21 Wh top-down, about 13 times at one boundary (glossary and Chapter 29), about nine times as first quoted in Wh at two boundaries (Chapter 20); the likely cause, idle capacity, is not measured and the Cloud A burst is unmetered |
Both, never merged, with direction | Meter a fortnight on the Frankfurt rack (AIDR-1379) |
| OT-HM-77 | Population of the claims assistant | About 320 handlers in the pilot unit (profile of AI Governance; Landscape reach L); AI Environment: about 2,900 active handlers and about 140 more in the UK and Switzerland (AIDR-1376) |
Pilot against rollout, both kept | Date each figure |
| OT-HM-80 | The same number twice | EBG-HM-0004 9,500 kWh a month (facility boundary) and a Cloud A batch of about 9,500 accelerator hours; also the 9,500 employees |
Different things | Change one figure |
| OT-HM-81 | Stop-drill records of Book 14 | The mandate gives the passing drill as 2.2 minutes and the text as 2 minutes 10 seconds (rounding); the stop path of EBG-HM-0002 lies in the claims application, that of EBG-HM-0001 in the gateway and a claims-core flag |
Text value | Align the record |
| OT-HM-87 | Wording of DORA's management body | Earlier: the executive board is the management body; Landscape now: the management body of a financial entity, which Article 5(2) makes responsible for its ICT risk management framework, is the board of the decision tiers (T4); an insurer within the scope of Article 2(1) | Executive board in Hamburg, with Article 5(2) | Check at entity level for the other EU entities |
| OT-HM-89 | Roles unnamed in Book 14 | Head of claims operations (Book 13 names Joanna Kaczmarek), head of claims operations at the Polish centre, head of underwriting; owners of BEN-HM-0001, BEN-HM-0002 and FPR-HM-0004 |
Roles as written | Name the holders |
(3, Nordqvist, 2026) (6, Nordqvist, 2026) (7, Nordqvist, 2026) (9, Nordqvist, 2026) (10, Nordqvist, 2026) (11, Nordqvist, 2026) (13, Nordqvist, 2026) (15, Nordqvist, 2026) (Tools, .) (16, Nordqvist, 2026) (14, Nordqvist, 2026) (15, Nordqvist, 2026)
What the register shows
Eighty-nine threads are recorded: 38 contradictions, 9 identifier collisions, 13 date clashes, 15 missing outcomes and 14 scope mismatches. The fifteen threads OT-HM-75 to OT-HM-89 come from AI Environment (Book 14) and from the revision of Landscape; the revision resolved none of the earlier threads. Four observations are the dossier's own conclusions.
First, the threads cluster on a few subjects and not across the profile. The agent count, the level of HM-SEC-001, the stop drill, the token life and the incident identifiers appear in four to six volumes each. The profile of the organisation (country count, headcount, supervisory frame, the shape of the boards) is almost free of conflict. A reader who wants the whole picture can therefore trust the profile and should treat the agent estate as the place where the series has not yet settled its own count.
Second, most clashes arise from the series order. A volume cannot know what a later volume will say, so the later volume re-dates, renames or recounts, and the earlier is never revised. The corrections the ledger already applies (the findings of AI Purple, the AIDR numbers of AI Boss) show the cheaper route: when a later volume is written, the earlier chapters it affects are corrected in the same set of changes.
Third, missing outcomes are rarely errors. A volume ends with a review date, a drill or a plan, and the next volume does not return to it. They matter because the series' own rule is that a claim carries what was tested, when and with what result; a stated test without a result is a claim with no evidence, and its age grows every month (Chapter 23).
Fourth, three threads are of a different weight: OT-HM-01, OT-HM-02 and OT-HM-33. They change what a reader concludes about how far the group trusts its agents and how many incidents it has had. The recommended order for the author is those three first, then the contradictions on levels and ceilings (OT-HM-07 to OT-HM-11, OT-HM-15 to OT-HM-19), then the identifier families, then the dates, and last the outcomes, which can be written as new facts in the next volume.
Established values are those of the profile to 16 November 2026. Every entry about 2027 and later is a finding about illustrative fiction, and a recommended resolution is an editorial choice for the author, not a correction of fact (2, Nordqvist, 2026).
Summary
The register holds 89 open threads in five kinds. The agent estate is where the volumes disagree most; the organisation's profile is where they agree. The dossier used a working value wherever a chapter needed one, never silently, and recorded it here with a recommendation. The chapter's own conclusion is that the series is consistent about who decides and inconsistent about how many, how old and under which identifier. These are inexpensive faults to fix, and the order in which to fix them is the order of the register.
Sources
- Nordqvist, G. AI Governance: Governing Artificial Intelligence in European Enterprises and Public Authorities. (2, Nordqvist, 2026)
- Nordqvist, G. AI Platform and Security: Building and Securing AI Platforms in European Enterprises and Public Authorities. (3, Nordqvist, 2026)
- Nordqvist, G. AI Red Team: AI as an Active Part of Cyber Defence in European Enterprises and Public Authorities. (4, Nordqvist, 2026)
- Nordqvist, G. AI Threat: Are the Dystopias Coming True? Evidence, Risk and Response for European Enterprises and Public Authorities. (5, Nordqvist, 2026)
- Nordqvist, G. The Transformed Security Architect: Role, Craft and Career in the Age of AI in European Enterprises and Public Authorities. (6, Nordqvist, 2026)
- Nordqvist, G. AI Boss: When Artificial Intelligence Manages People in European Enterprises and Public Authorities. (7, Nordqvist, 2026)
- Nordqvist, G. AI Blue: How Artificial Intelligence Transforms the Blue Team in European Enterprises and Public Authorities. (8, Nordqvist, 2026)
- Nordqvist, G. AI Purple: How Red and Blue Learn Together When AI Joins the Loop in European Enterprises and Public Authorities. (9, Nordqvist, 2026)
- Nordqvist, G. AI Operations: How Artificial Intelligence Transforms IT Operations in European Enterprises and Public Authorities. (10, Nordqvist, 2026)
- Nordqvist, G. AI Leader: How Leadership Roles Change When AI Is Used Broadly in European Enterprises and Public Authorities. (11, Nordqvist, 2026)
- Nordqvist, G. AI Developer: How the Developer Role Changes When AI Is Used Broadly in European Enterprises and Public Authorities. (12, Nordqvist, 2026)
- Nordqvist, G. Vibe to Dev: From Vibe Coding to Professional Software, Step by Step, for Builders Who Are Not Developers, in European Enterprises and Public Authorities. (13, Nordqvist, 2026)
- Nordqvist, G. AI Environment: Designing AI for the Greatest Benefit and the Least Environmental Harm in European Enterprises and Public Authorities. (14, Nordqvist, 2026)
- Nordqvist, G. Landscape: A Fact-Based Picture of Architecture and AI at Work, for Leaders, Architects and Citizens. (15, Nordqvist, 2026)
- Nordqvist, G. Tools: A Reference to the Templates, Software and Methods of The Transformed Architecture for European Enterprises and Public Authorities. (16, Nordqvist, 2026)